Roland Jet Phaser Part substitutions

[joelap]

Hey all,

Going off of this schematic I found by searching these glorious forums:


Have a couple of questions:
1) Bottom right in the bipolar power supple section, what are those two caps to the left of the 9V batteries?  Would I be right in assuming those are filtering caps for the power supply, and therfore arent "critical" that the values be exact to the original?  What would be some good values, would 470uf electrolytics work?  I've got a bunch of those lying around.
2) I'm looking at "All FETs are 2SK30A(Y) Matched".  Well the 2SK30A is an N-Channel JFet... could I just substitute that part for the cheap and plentiful J201's?  Buildyourownclone used to sell matched FETs but it seems they've removed the parts section of their site... any other place selling matched FETs suitable for this project?
3) What opamps are suitable for this project? 

Thanks.  If I can get this to materialize, I'll post a board layout for everyone.

[joelap]

Anyone?

[Mark Hammer]

K30 FETs are available.  The amount of time it takes to match FETs tends to warrant against distributors doing the matching for quite that many.  You can buy the requisite FETs at Smallbear, though.  J201s are not kown for their sweep properties in phasers.  I'd save mine for preamps if I was you.  MPF102 and 2N5457 might be better choices if you can't find 2SK30A in sufficient quantity.

The op-amps can really be just about anything, though I'd recommend a TL022 for the LFO section.

[joelap]

Thanks Mark.  Now that leads me to two questions:

How is the consistency among those 2SK30A FETs?  Would I be safe buying 20 and be able to get 8 within range, or are they more variable than that?

And what about those power supply caps?

[Bill Bergman]

I made a pcb layout for this if your interested. It's a really cool phaser.
...
Bill

[puzzle87]

look out! probable error in schematic:

[Bill Bergman]

Here's a layout I put in the gallery a while back... http://aronnelson.com/gallery/3dm

[moosapotamus]

Nice to see you, Bill.

~ Charlie

[Bill Bergman]

Thanks Charlie... I'm still alive....but ... am I suppost to be.... is that the question? :

[oldschoolanalog]

Nice layout. I have to ask the obvious: Is there component placement/value info for this board somewhere? BTW, Mr. Bergman; have you checked out the 19 page A/DA flanger megathread?  You were a big inspiration for lots of folks who aspire to take on this project. I think that thread will give you a  .

[moosapotamus]

Bill - the answer is... yes.

Isn't that layout a nicely cleaned up version of what's in the AP7 service notes?
I think there might be an overlay in there, too... maybe?

I'd love to build one of these, too... someday.

~ Charlie

[Mark Hammer]

How is the consistency among those 2SK30A FETs?  Would I be safe buying 20 and be able to get 8 within range, or are they more variable than that?

That's an interesting and thought-provoking question.  Knee-jerk response would say that they all need to be matched, and that you'd probably need about 30 or more to do that.

On the other hand.....

The reason you want to match FETs is so that they all sweep uniformly.  When FETs are unmatched, it can be the case that, for some portion of the LFO sweep range, three of the FETs are still happily changing their resistance, but one of the FETs declares "I can't do any more guys.  Carry on without me.", and stops changing resistance no matter how much farther the LFO sweeps in that direction.  Let me note that it is not the absolute resistance of those individual FETs, but the fact that some are changing in resistance, and some are not, because they respond to gate voltages differently.

But let's consider the impact of such mismatching in different contexts.  If one FET fails to track the other in a 2-stage phaser, I lose any hope of creating a moving notch.  If one or two stop changing in a 4-stage unit, I can drop back from 2 notches to one at the extremes of the sweep; a noticeable drop in "effect intensity".  So what happens if a couple of FETs don't quite go the distance in an 8-stage unit?  Will the perceived intensity of the effect be diminished?  Like I say, interesting question.  My gut sense is that dropping back from 4 to 3 notches will not be nearly as noticeable as dropping from 2 to 1, or 1 to none.  So, in one sense, one might predict there is greater tolerance for FET mismatching in more complex phasers.

On the other hand, the more FETs there are to mismatch, the greater the likelihood that mismatch could occur at each end of the sweep, rather than just at one end.

In some respects, this is an empirical question, that can be easily resolved by taking an 8-stage unit and subbing one, then 2, then 3 randomly selected FETs and subbing the matched units with these other FETs to see at what point mismatching starts to make a diference.

Of course, it might present an argument for adopting looser matching criteria for the more complex units.

Still, I'm curious either way.

[Bill Bergman]

I used BF244A 's. I can't remember how well or if the were match. I used a surface mount pot instead of a trimpot. This allows you to tune the frequency of the sweep.

[Bill Bergman]

I'd post more pics but I don't remember how.

[mdh]

From an *extremely* cursory look at the data sheets, I suspect that 2N5485s would work well in this circuit. This is the part that R.G. typically recommends for MXR Phase 90 clones and derivatives, and this circuit looks pretty similar (though I could be missing something important... experts, please chime in if I am). I bought a lot of 50 Fairchild 2N5485s on eBay (about $7 US) recently for my Phase 90, hoping to get a closely matched quartet. In fact, I ended up with a few quartets that matched perfectly within the precision of my DMM, and at least one octet that matched within 1 or 2 mV. So I can't speak for the 2SK30A, but a pile of 50 2N5485s would probably do, if they really are comparable in spec to the 2SK30A.

[joelap]

From an *extremely* cursory look at the data sheets, I suspect that 2N5485s would work well in this circuit. This is the part that R.G. typically recommends for MXR Phase 90 clones and derivatives, and this circuit looks pretty similar (though I could be missing something important... experts, please chime in if I am). I bought a lot of 50 Fairchild 2N5485s on eBay (about $7 US) recently for my Phase 90, hoping to get a closely matched quartet. In fact, I ended up with a few quartets that matched perfectly within the precision of my DMM, and at least one octet that matched within 1 or 2 mV. So I can't speak for the 2SK30A, but a pile of 50 2N5485s would probably do, if they really are comparable in spec to the 2SK30A.

Nice.  I think I'll get started on this after the holidays, and I'll grab a bunch of both and see how many I can get to match fairly well, maybe I'll try both the 2N5485 and the 2SK30A's.

Still wondering about those caps to the left of the power supply... are their values critical that they are exact?  And is puzzle87 correct about there being an error there in the schem?